Rigid-flex PCB design
When there is no room for connectors and no tolerance for a harness that shakes loose.
Rigid-flex is what you reach for when the mechanics have run out of room and the environment has run out of patience.
In a cylindrical body, a sealed enclosure, or anything that gets shaken hard, board-to-board connectors and wire harnesses are the two most common failure points. Rigid-flex deletes both: the interconnect becomes part of the board, fabricated once, with no mating cycle to degrade and nothing to work loose.
Where the engineering actually is
The bend, not the board. The rigid sections are ordinary PCB design. The flex regions and — especially — the transitions between them are where a rigid-flex design succeeds or gets scrapped. Copper in a bend zone wants to be continuous, hatched rather than solid, running perpendicular to the fold, and nowhere near a via.
Bend radius against the real stack-up. Calculated from finished thickness and copper weight, distinguishing a one-time installation fold from a dynamically flexing region. These have very different limits and conflating them is the classic first-time mistake.
Stack-up transitions. The layer structure changes where rigid meets flex, and the fabricator has firm opinions about how. Coverlay openings, stiffener placement and the keep-out around the transition all get settled with them before layout, not after.
Impedance across two geometries. A differential pair crossing a flex region has to hold its impedance in both stack-ups, with no discontinuity at the join. Achievable, and it has to be planned.
Fold order. A rigid-flex assembly is folded by a person, in a sequence, into an enclosure. If that sequence is impossible or ambiguous, the design fails at assembly regardless of how well it routes. It gets documented as a deliverable.
Real-world example
For a tube-launched fixed-wing UAV, the entire avionics assembly had to fit inside the body diameter and survive launch loads. Three rigid sections joined by flex, manufactured as a single assembly at Eurocircuits, with bend-radius calculation and stack-up transitions worked through with the fabricator before layout began.
No board-to-board connectors, no harness, nothing to unseat under shock — in a volume where none of those would have fitted anyway.
When we will tell you not to
Rigid-flex costs more per board, has longer lead times, and is unforgiving of late changes. If your enclosure has room and your product sits still, a connector and a ribbon cable is the better engineering decision and we will say so.
Common questions
Straight answers
When is rigid-flex worth the cost?
When the assembly it replaces costs more than the board. Rigid-flex removes connectors, harnesses, mating operations and inspection steps — each of which is a part cost, a labour cost and a failure point. In a tight enclosure, or anywhere subject to shock and vibration, it frequently costs more per board and less per product. In a roomy box with a low-vibration life, it usually is not worth it.
What bend radius can we actually achieve?
It depends on the finished flex thickness, copper weight and whether the bend is one-time (installation fold) or dynamic (repeated flexing). A common rule of thumb is ten times the flex thickness for a static bend and considerably more for dynamic. We calculate it against the real stack-up rather than a rule of thumb, and confirm it with the fabricator before layout — because a bend radius that fails is a scrapped panel, not a rework.
Can signals keep their impedance across a bend?
Yes, with care. The flex region has a different stack-up from the rigid sections, so the impedance target has to be met twice with different geometry, and the transition itself must not introduce a discontinuity. High-speed pairs crossing a flex are entirely possible and need to be planned rather than routed hopefully.
Which fabricators do you design to?
We have built three-board single rigid-flex assemblies at Eurocircuits and design to whichever EU fabricator you prefer. Their capability sheet becomes the design rule set before layout starts — rigid-flex is a process where a generic rule set produces an unmanufacturable board.
What about semi-flex?
Semi-flex — a thinned FR-4 region rather than a true polyimide flex — is cheaper and suits a single installation fold with a generous radius. It is the right answer more often than people expect, and the wrong one for anything dynamic. We will tell you which case you are in.